Numerical simulation of tomography-SAR imaging and the object reconstruction using the compressive sensing approach with L1/2-norm regularization

Making use of multiple acquisitions of the SAR observations over the same area, the tomography-SAR (tomo-SAR) technology can achieve a three-dimensional (3-D) imagery of the interested scene. Given the sparseness of the elevation signals, the compressive sensing (CS) approach has become an effective and innovative method. Some regularization tools are also employed in CS approach to reconstruct the reflectivity profile of the objects. In this paper, we present a novel L_1/2-norm regularization to realize 3-D reconstruction. As a non-convex optimization problem, the L_1/2-norm regularization is solved by an iterative algorithm. Multiple acquisitions might cost expensive big data, e.g. extending one or more years. Based on our studies on polarimetric scattering and SAR image simulation, we first apply numerical simulation of polarimetric scattering for multi-pass SAR observations and produce the tomo-SAR image of the terrain objects. It can be of great help for parameterized analysis of tomo-SAR imagery. As an example, the tomo-SAR image and 3-D reconstruction of the Beijing National Stadium is presented.